Fast Power Control Adaptation via Meta-Learning for Random Edge Graph\n Neural Networks

2021/05/02 by Ivana Nikoloska, Osvaldo Simeone, Nikoloska, Ivana +1
Computer Science · Engineering · #Advanced MIMO Systems Optimization #Cooperative Communication and Network Coding #Energy Harvesting in Wireless Networks #FOS: Computer and information sciences #FOS: Electrical engineering #Information Theory (cs.IT) #Machine Learning (cs.LG) #Signal Processing (eess.SP) #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2105.00459

openalex publication_date 2021/05/02 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Power control in decentralized wireless networks poses a complex stochastic\noptimization problem when formulated as the maximization of the average sum\nrate for arbitrary interference graphs. Recent work has introduced data-driven\ndesign methods that leverage graph neural network (GNN) to efficiently\nparametrize the power control policy mapping channel state information (CSI) to\nthe power vector. The specific GNN architecture, known as random edge GNN\n(REGNN), defines a non-linear graph convolutional architecture whose spatial\nweights are tied to the channel coefficients, enabling a direct adaption to\nchannel conditions. This paper studies the higher-level problem of enabling\nfast adaption of the power control policy to time-varying topologies. To this\nend, we apply first-order meta-learning on data from multiple topologies with\nthe aim of optimizing for a few-shot adaptation to new network configurations.\n

Related

• Modular Meta-Learning for Power Control via Random Edge Graph Neural Networks
• A Graph Neural Network Approach for Scalable Wireless Power Control
• Wireless Power Control via Counterfactual Optimization of Graph Neural Networks
• Meta-Learning Empowered Graph Neural Networks for Radio Resource Management
• Multi-Agent Reinforcement Learning for Power Control in Wireless Networks via Adaptive Graphs